The current additive manufacturing (AM) product development environment is far from being mature. Both software applications and workflow management tools are very limited due to the lack of knowledge supporting engineering decision making. AM knowledge includes design rules, operation guidance, and predictive models, etc., which play a critical role in the development of AM products, from the selection of a process and material, lattice and support structure design, process parameter optimization to in-situ process control, part qualification and even material development. At the same time, massive AM simulation and experimental data sets are being accumulated, stored, and processed by the AM community. This paper proposes a four-tier framework for self-improving additive manufacturing knowledge management, which defines two processes: bottom-up data-driven knowledge engineering and top-down goal-oriented active data generation. The processes are running in parallel and connected by users, therefore forming a closed loop, through which AM knowledge can evolve continuously and in an automated way.
